Pathway

Cytochrome P450s (CYPS)

Strategies for debottlenecking

This shows Table 1 review by 2020.

Shows studies grouped into strategies

Point is. No general strategy. (Depends on particular bug in system. Hard to troubleshoot)

with this in mind (time and resources) we made following hypotheses

Skip details Highlighted highest level.

1. Expres: N-terminal could be something

2. Protein engineering: typically works but need HT screening (functional homologs)

3. Coexp: Fusions CPRs needs to service multiple P450's

4. Chassis engineering: May introduce more problems. Growth.

Main hypothesis

Specific combinations of CPR / G8H homologs and corresponding expression levels can remove the G8H bottleneck in the Strictosidine pathway

Experimental setup:

Results

Results

many more producers
but not much better than last round.
plateu indicate another bottleneck

Discussion

Feeding showed SLS bottleneck
Create SLS library.
But that’s up to you guys

Strategy

First point - If we new exactly how cells work we could improve by creating a single genotype. - When solution is unknown. Success correlates with no of tests/genotypes - Combinatorial libraries create many genotypes - we create libraries in every step! We could have chosen to selectively create ML predicted genotypes.

Second point - here we used two strategies: - explorative (sampled few in large library) - exploitative (used ML to recommend optimal parts within a library) - switching between these strategies seems to work. - When no jumps in fitness i.e. plateu, explore. - When jumps in fitness ie. expo curve, exploit.

Where we are in laboratory
talk about improving workflows with teemi

Teemi

This figure tries to capture Teemi. Lets work through it.

This is natural… a normal protocol (like from lab or kitchen)

Its written in ENG.

That can’t be improved? Anything wrong with english?

Two things

Ambiguous (Almost to powerful, it can describe everything, in many different ways and on many levels. Who hasn’t tried follow instruction and found it needs details and is unclear!)
After reading, we have to perform the actions!

This is literate… natural language and coding language protocol, e.g. ENG and Python

We have converted by translating lines from ENG to CODE

What’s the benefit.

Code unambiguous (We all know coding is a way to describe how to do things e.g. add numbers, … and it does it unambigously)
If description succeeds, Computers can perform for us. in principle everything can be done by a computer or computer connected to a robot

Nice properties, for automation, reproducibility, error reduction, and knowledge transfer etc.

Teemi is:

Example of documenting lab workflows in literate programming style.
a python package with functions etc. that can be used across synbio projects.

In the middle, circle, DBTL cycle, natural way break down synbio projects, teemi covers all steps.

Sounds great, but can you show us some examples?

Project parts

Design
- Promoter & homolog selection
Build
- Background strain
- Library
Test
- Library characterization
Learn
- Plotting and modelling

Project notebooks

1. Organise samples

2. Simulate experiments

3. Plan reactions

Summary

Benefits of literate programming (as I see them)

One system
Automate everything repetitive
Error reduction
Next level documentation
Resist knowledge loss

Ultimate goal: more time for biological questions

One system (jupyter notebook[+ python packages]). With literate programming all of a scientists work can be documented (DBTL):D (DNA design) B (laboratory protocols) T (preprocessing of data) L (plotting and modelling). Notebooks can also control other programs CLC (construct planning). IDT.com (ordering primers), Teselagen/Benchling (LIMS / sample tracking), Excel (calculating mastermixes), (R dataanalysis) with executable lines. Huge potential even before including robots/automation. This can only be done by collaboration between those who understand IT/automation and those that are close to the biological problems.open-ended: Flexible tool. write programs that do whatever we want, our own: We have control. No payment or support taken away
Allows us to automate everything repetitive.
Error reduction. a) unambiguous b) Also simulate before execution
Next level documentation. Unambiguous Executable instructions. Increase citations because more useful to others
Our business vulnerable to knowledge loss (rapid student/employee turnover). Digitalisation will oppose this. Personal experience: worked 5 years in lab. I cannot remember details from time to time. Tasks can be very complex errors occur if not systematic need to retain efficiently. Also great help when needing to pass on knowledge to someone else.

programming just a tool. Ultimate goal: more time for answering biological questions

Other points: - Continually maximisese work of computers. Such that minimise our routine work and maximise our focus on biological questions - Python circumvent the need to learn and switch between 10+ SPECIALIZED applications that doesn’t talk together. Also if something doesn’t do what you want, you yourself, can change it immediately. - Project outcome: The good example: Good examples of jupyter notebooks and python packages that will provide so many advantages that anyone want to incorporate in their work. - No one HAS to learn any python. We use it because it is the best. My prediction is that others will WANT to learn python and jyputer as they understand the benefits. Just as we did. Also doesn’t need to learn to make own functionality (although I predict they will quickly want to). Just use others who can and will collaborate.

Where does abstractions fit in? hide/controlling/suppressing complexity in funcitons?

Thank you for your attention!

Debottlenecking the strictosidine pathway - and improving workflows with teemi

By Søren Petersen

9 October 2024

Pathway

Cytochrome P450s (CYPS)

Strategies for debottlenecking

Main hypothesis

Experimental setup:

Results

Results

Teemi

Project parts

Project notebooks

Project notebooks

1. Organise samples

2. Simulate experiments

3. Plan reactions

Summary

Thank you for your attention!